基于改进BP算法的热电阻特性逼近与测温虚拟仪器设计

Probe into Thermal Resistance Characteristic Approximating Based on Improved BP Algorithm and Design of VI for Measuring Temperature

分析了热电阻的非线性温度特性,用改进BP神经网络算法建立了1个Pt100热电阻温度测量模型。并将数据库进行合理地归一化处理。讨论了神经网络的隐藏层数目、误差的选择、各种改进算法对本课题的合适程度,并用分度表严格训练网络,运行测试。用Labview软件设计虚拟仪器,使用MATLAB Script函数节点使之可以进行神经网络运算,此虚拟仪器用来直观地显示测试结果结果表明,使用改进BP算法神经网络对热电阻温度进行测量,可以明显提高测量精度。

This paper analyzes the nonlinear characteristics of resistance temperature, and establishes a temperature measurement model of Pt100 thermal resistor by using BP neural network. Meanwhile, the paper carries out the normalization of the database reasonably, calculates the number of hidden layer, discusses the selection of errors and various improved algorithms to the subject, makes the network training based on reference tables, designs VI to show the test resuh using Labview, and builds BP model with MATLAB Script. The results show that the model proposed in this paper carl improve the accuracy of the temperature measurement for the resistance temperature.